Structural and Mechanical Properties of CrN Thin Films Deposited on Si Substrate by Using Magnetron Techniques

Round 1

Reviewer 1 Report

The paper reports on the fabrication of CrN films on Si by sputtering. In particular, both DC and pulsed sputtering fields are used and three different deposition durations are used. Films are analyzed by SEM, EDAX, AFM (including stiffness and nanoindentation). 

Overall, the paper can be expected to provide with minimal contributions to the field of film deposition, since it reports average quality and standard characterizations with few, or no, discussion and explores only one process parameter (duration) of the adopted methods.

Nonetheless, it can be considered for publication, but Authors must account for the following points.

1.     Marker size in Figs. 1 and 2 must be made larger, in order to be read.

2.     I could not understand in which point are EDAX spectra in Figs. 3 and 4 acquired. If they are made over the whole area imaged in Figs. 1 and 2, they are completely unuseful. Otherwise, if they are acquired somewhere in the film region, Authors must explain the strong presence of Si signals. Moreover, figures are very difficult to be read because of small lettering size. Since they do not convey much information, Authors can also consider removing them from the paper.

3.     The elemental comment inferred from EDAX spectra and reported in Table 2 is also completely unclear to me. Once more, Authors must specify where EDAX data have been acquired. Moreover, the discussion at lines 157 and following seems to disregard a rather obvious consequence of using different deposition durations, namely the occurrence of films with thickness increasing along with the duration. Further to deeply revise all the EDAX part, Authors must comment on the thickness (which, by the way, appears to have been measured based on SEM cross-sectional images).

4.     SEM plane views in Figs. 5 and 6 are not very convincing: marker size must be made readable, but, in any case, the adopted field of view is definitely too large to enable any reasonable conclusion. Contrast is in fact very poor and personally I could not appreciate any relevant detail, nor remarkable differences between different samples. Also in this case, Authors may consider removing SEM images, unless they can provide better quality figures.

5.     Error bars must be provided in the thickness measurements of Table 5. In some case, thickness is expressed with an accuracy of 10 pm, which is absolutely not reasonable (and also much smaller than the resolving power of any electron microscope!). I can’t believe that thickness remains uniform (within such an incredible accuracy) over any length. Authors must repeat the thickness measurement along some line, determine the average value and take the standard deviation as the uncertainty of the measurement.

6.     It looks like AFM investigation and SEM images do not show any appreciable similarity. Since they are both showing film morphology, Authors must carefully comment on that.

7.     How can Authors state that defects of the Si substrate are visible in the film morphology (line 269)? Are they sure such defects come from the substrate? Usually, surface defects in standard quality Si substrates are hard to be detected with conventional techniques. Authors must carefully consider this point.

8.     No interpretation is given to the decrease of the Young’s modulus as a function of the film thickness (Table 6). In my opinion this is simply a consequence of the AFM tip indenting the whole film thickness, hence the reported Young’s modulus is the one of the (stiff) Si substrate. Authors must include appropriate comments on that, possibly referring to the wide literature on the topic of nanoindentation of thin films.

9.     Authors report R_q thickness, which is a line-based measurement, rather than S_q, the surface areal roughness. I think it would be more appropriate to use S_q rather than R_q.

10.  A comparison of the fractal dimension (FD) obtained via AFM and SEM should be reported in the paper.

11.  Language and style should be carefully revised. At least, Silicone and places at line 59 must be corrected (into Silicon and placed, respectively).    

Author Response

We address many thanks to the Reviewer for their thoughtful and professional  insights in respect to our manuscript. For an easy-to-read answer, we will respond point-by point in the following. Their comments are left as they were, while our responses are inserted correspondingly with green color. 

1. Marker size in Figs. 1 and 2 must be made larger, in order to be read.

According to the advice, the marker size in Figs. 1 and 2 were made larger.

2. I could not understand in which point are EDAX spectra in Figs. 3 and 4 acquired. If they are made over the whole area imaged in Figs. 1 and 2, they are completely unuseful. Otherwise, if they are acquired somewhere in the film region, Authors must explain the strong presence of Si signals. Moreover, figures are very difficult to be read because of small lettering size. Since they do not convey much information, Authors can also consider removing them from the paper.

The EDAX spectra were acquired in the cross section of the films, inside the region of the red boxes delimited in Figs. 1 and 2. The EDAX spectra investigations had two purposes: Firstly, to achieve the EDAX spectra, which qualitatively confirms the presence of Cr and N elements after deposition and assures that no contamination occurred during deposition process. And secondly, to verify the transversal distribution of chemical elements and its correlation to the SEM transversal images. Therefore, we consider these figures worth to be included in the paper.

The size of the electron beam is larger than the CrN layer which is very thin (between 50-400 nm depending on the technique as well as the deposition time), so that it penetrates the Si substrate, which explains the strong presence of Si signals.

The lettering size was increased to increase the readability of the figures.

3. The elemental comment inferred from EDAX spectra and reported in Table 2 is also completely unclear to me. Once more, Authors must specify where EDAX data have been acquired. Moreover, the discussion at lines 157 and following seems to disregard a rather obvious consequence of using different deposition durations, namely the occurrence of films with thickness increasing along with the duration. Further to deeply revise all the EDAX part, Authors must comment on the thickness (which, by the way, appears to have been measured based on SEM cross-sectional images).

The EDAX data was acquired in the area inside the red boxes from Figs. 1 and 2. The EDAX analysis confirms that the deposition process was successful with the existence of Cr and N signals in Figs. 3 and 4 and shows that the deposited layer is homogeneous and uniform with no noticeable defects or contamination in the deposition layer (Figs. 1 and 2). To eliminate any confusion we removed Table 2 and left only Figs 3 and 4 that certify the existence of the CrN layer.

4. SEM plane views in Figs. 5 and 6 are not very convincing: marker size must be made readable, but, in any case, the adopted field of view is definitely too large to enable any reasonable conclusion. Contrast is in fact very poor and personally I could not appreciate any relevant detail, nor remarkable differences between different samples. Also in this case, Authors may consider removing SEM images, unless they can provide better quality figures.

The SEM images in Figs. 5 and 6 were replaced with better quality ones and the marker size was increased. In the new figures the behavior that we wished to capture is represented even better.

For Fig. 5 we wished to show that, with the increase on the deposition time, a fine and columnar growth structure forms on the deposited CrN layer. This structure is representative of the DC sputtering method and can also be seen in the cross-section view (Figure 7).

For Fig. 6 no growth pattern is noticeable because the deposited layer is dense and glassy (which again is representative of HiPIMS sputtering and seems to hold true even in our case of ultrathin depositions).

The manuscript was updated consequently with the above clarifications.

Additionally, we do not use the SEM image only for the visual appreciation of details, but we also differentiate the samples using quantitative measurements such as FD, Correlation and Energy.

5. Error bars must be provided in the thickness measurements of Table 5. In some case, thickness is expressed with an accuracy of 10 pm, which is absolutely not reasonable (and also much smaller than the resolving power of any electron microscope!). I can’t believe that thickness remains uniform (within such an incredible accuracy) over any length. Authors must repeat the thickness measurement along some line, determine the average value and take the standard deviation as the uncertainty of the measurement.

The reviewer is correct, and we appreciate their valuable suggestions. The thickness measurements were repeated along some lines and error bars were added to express the uncertainty of the measurements.

6. It looks like AFM investigation and SEM images do not show any appreciable similarity. Since they are both showing film morphology, Authors must carefully comment on that.

The AFM and SEM images do not show any appreciable similarity because they capture different characteristics of the ultrathin film. The two investigation techniques are complementary: AFM shows the film topography as quantitative dimensional measurement of features on the surface, while SEM shows the film morphology as shape, texture, and distribution of materials at the investigated surface.

7. How can Authors state that defects of the Si substrate are visible in the film morphology (line 269)? Are they sure such defects come from the substrate? Usually, surface defects in standard quality Si substrates are hard to be detected with conventional techniques. Authors must carefully consider this point.

The reviewer is correct, and we carefully revised the discussed paragraph. We replaced the word “defects” with “microscopic scratches”, and we clarified that the microscopic scratches of the Si substrate (due to sample handling before the deposition) were still visible after the deposition process.

8. No interpretation is given to the decrease of the Young’s modulus as a function of the film thickness (Table 6). In my opinion this is simply a consequence of the AFM tip indenting the whole film thickness, hence the reported Young’s modulus is the one of the (stiff) Si substrate. Authors must include appropriate comments on that, possibly referring to the wide literature on the topic of nanoindentation of thin films.

Indeed, the Young’s modulus decreases as a function of the film thickness. The reviewer is right regarding the fact that the tip indents the whole film thickness and reaches the Si substrate. However, given this possibility we decided to extract the Young’s modulus from force-distance curves, which is a technique that does not alter the surface of the samples, but is affected by the surface nanoscale particularities via van-der-Waals forces. For the case of DC power supply, the columnar deposition which is gradually formed with the increase of the time deposition strongly influences the tip-surface interaction. On the other hand, in the case of HiPIMS power supply, the films are very thin and consequently the tip-surface interaction is influenced by the substrate as well. As the film thickness increases, this influence weakens and the measured Young’s modulus decreases becoming close to the one of the CrN. 

To obtain a sense of the stiffness of the samples, we computed the stiffness from shallow nanoindentation, as descripted in the Methods section. Form these measurements, we observe that the stiffness increases with the film thickness, which is what we expected.

The manuscript was updated with the appropriate comments on topic.

9. A comparison of the fractal dimension (FD) obtained via AFM and SEM should be reported in the paper.

We appreciate the reviewer’s suggestion. However, the fractal dimension obtained via AFM and SEM cannot be easily compared since they were extracted from different types of measurements. They are discussed separately because they show different characteristics of the film, one shows the fractal dimension of the topography of the film while the other shows the fractal dimension of the morphology.

10. Language and style should be carefully revised. At least, Silicone and places at line 59 must be corrected (into Silicon and placed, respectively).  

The language and style have been carefully revised.

Reviewer 2 Report

In this work, Tranca and c-workers performed an investigation of the structural and mechanical properties of CrN thin films deposited on Si substrates. The films were deposited by magnetron techniques. Here, the authors indicate that they have found that structural/mechanical properties of thin films are affected by deposition time/method.

The topic of research is indeed a good fit for this journal. The manuscript is relatively well-written and in general appears to be reasonably scientifically sound. However, the study itself appears to be rather straightforward and  a bit simple. Nevertheless, there is some potential for acceptance if the the authors address the following in Major Revision:

-There is some inconsistencies in the definition (or not) of abbreviations. For example, in the first paragraph of Introduction, the authors define N as abbreviation of Nitrogen but do not define what Cr, BCC, Cr2N and CrN are abbreviations of;

-Line 50 of page 2, the authors indicate that CrN films where deposited by two techniques. Here, the authors should properly identify and characterize them, and justify why they were tested;

-Why the chosen deposition times?

-The labels of Figure 3 are quite difficult to read;

-There is no proper analysis of the results of Figure 1-4, as the authors just present the Figures without describing/discussing the obtained results;

-While the authors determine/evaluate different parameters and some correlation is made with deposition method/time, there is no proper discussion about how these findings can be used to advance the field of CrN thin films;

-How do the obtained results compare with existent literature?

-Why the choice of mainly using only two characterization techniques (AFM and SEM)?

-How common is the use of the deposition method/time here employed for the fabrication of CrN films?

-The authors conclude that deposition time and method play a crucial role in the final structure and mechanical properties of the deposited layers. Is this not a conclusion already found in the literature?

Author Response

We address many thanks to the Reviewer for their thoughtful and professional insights in respect to our manuscript. For an easy-to-read answer, we will respond point-by point in the following. Their comments are left as they were, while our responses are inserted correspondingly with green color. 

-There is some inconsistencies in the definition (or not) of abbreviations. For example, in the first paragraph of Introduction, the authors define N as abbreviation of Nitrogen but do not define what Cr, BCC, Cr2N and CrN are abbreviations of;

We appreciate the observations of the Reviewer. The noticed inconsistencies have been resolved.

-Line 50 of page 2, the authors indicate that CrN films where deposited by two techniques. Here, the authors should properly identify and characterize them, and justify why they were tested;

The techniques used in this paper are both methods of physical vapor deposition. The main difference comes from the type of power supply used. The first method, direct current sputtering (DC), uses a direct current power supply while the second method, high-power impulse magnetron sputtering (HiPIMS) uses an alternating current power supply. We used these techniques since both are popular in industrial applications. DC sputtering is used for coating the edge of cutting tools and HiPIMS sputtering is used for the coating of electronic circuits that also act as structural elements.

The justification of the tested methods was inserted in the manuscript.

-Why the chosen deposition times?

The deposition times of 5, 10 and 15 minutes were chosen because we wanted the deposited layers to be ultrathin (below 400nm). During our experiments we tested different deposition times and noticed that for deposition times greater than 15 minutes the thickness of the deposited layer would exceed 400 nm. In industrial application the target starting thickness of the deposited layer is in the order of micrometers, and consequently the starting deposition time is usually 180 minutes. But we wanted to study ultrathin layers less than 100 nm which is an unexplored field.

-The labels of Figure 3 are quite difficult to read;

The label size in all figures was increased to have better readability.

-There is no proper analysis of the results of Figure 1-4, as the authors just present the Figures without describing/discussing the obtained results;

Figures 1 and 2 have been used for the interpretation of the homogeneity and for the uniformity of the deposited layers. Figures 3 and 4 have been used for chemical analysis of the CrN thin films deposited on the Si substrate. We added proper discussions in the manuscript.

-While the authors determine/evaluate different parameters and some correlation is made with deposition method/time, there is no proper discussion about how these findings can be used to advance the field of CrN thin films;

What is new about this study is the characterization of DC and HiPIMS sputtered deposition that are on the order of tenths of nanometers. There are no reported results in literature for such thin films of CrN deposited on Si substrates using the two deposition methods presented in this paper. Additionally, no paper explores the quantitative characterization by fractal dimension, correlation, and energy. A short discussion on this topic is inserted in the manuscript.

-How do the obtained results compare with existent literature?

There are no existing results in literature for such thin films (50-400 nm) of CrN deposited on Si substrates using the two deposition methods presented in this paper. Particularly no paper explores the quantitative characterization by fractal dimension, correlation, and energy.

-Why the choice of mainly using only two characterization techniques (AFM and SEM)?

The two characterization techniques were chosen due to their popularity in the field of thin films and coatings.

-How common is the use of the deposition method/time here employed for the fabrication of CrN films?

DC and HiPIMS sputtering are two of the most popular techniques used in the fabrication of CrN films because they are inexpensive and can produce many samples in a short amount of time.

In industrial applications the deposition times of 5,10 and 15 minutes are not used  of, mainly because most industrial application require coatings in the range of micrometers to hundreds of micrometers not tenths to hundreds of nanometers.

-The authors conclude that deposition time and method play a crucial role in the final structure and mechanical properties of the deposited layers. Is this not a conclusion already found in the literature?

The reviewer is correct, such findings are already found in literature but not for such small deposition times. Besides, to our knowledge, no other paper explores the quantitative characterization by fractal dimension, correlation, and energy of CrN films deposited on Si substrate with the two methods presented in the paper.

Reviewer 3 Report

This work reports on Cr-N deposition using magnetron sputtering techniques. They have obtained the best deposition technique considering mechanical and structural properties. It is claimed to be a highly protective coating and can be used for PV devices. This report is interesting. It can be considered for publication after major revision.

Comments:

1.    The CC images in Fig. 7 and 8 do look not good. It is recommended to revise the images with identical appearance/thickness of Si-substrate. So that we can clearly see the variation of CrN thickness only.

2.    Why we could not see fine SEM images; as reported here; https://www.jstage.jst.go.jp/article/isijinternational/62/1/62_ISIJINT-2021-318/_pdf. If possible, it is suggested to re-check it.

3.    Please include XRD results to evaluate the crystallinity of CrN films grown in different sputtering methods.

4.    Since the authors have claimed to be used as a protective layer in PV devices. It is recommended to include a water contact angle (as reported; https://www.sciencedirect.com/science/article/abs/pii/S0040609022003984) on the sputtered film. It is suggested to discuss the surface properties compared with other kinds of sputtered films.

5. It is suggested to include an error bar in the tabulated results wherever it applies. 

Author Response

       We address many thanks to the Reviewer for their thoughtful and professional insights in respect to our manuscript. For an easy-to-read answer, we will respond point-by point in the following. Their comments are left as they were, while our responses are inserted correspondingly with green color. 

1. The CC images in Fig. 7 and 8 do look not good. It is recommended to revise the images with identical appearance/thickness of Si-substrate. So that we can clearly see the variation of CrN thickness only.

According to the reviewer’s suggestion, the images in Figs. 7 and 8 have been revised and the variation of the CrN thickness is clearly seen.

2. Why we could not see fine SEM images; as reported here; https://www.jstage.jst.go.jp/article/isijinternational/62/1/62_ISIJINT-2021-318/_pdf. If possible, it is suggested to re-check it.

We have added other SEM images with 400000x magnification that are clearer. The SEM images are still not as fine as the ones reported in that paper because we use a low accelerating voltage of 20 kV because at larger voltages the CrN film surface could be affected by collisions with fast electrons.

3. Please include XRD results to evaluate the crystallinity of CrN films grown in different sputtering methods.

Thank you for the great suggested idea. But we do not plan to include XRD results in this paper. We have another manuscript in preparation that studies the interface of the ultrathin CrN deposition with the Si substrate in which we plan to include film crystallinity obtained by XRD measurements.

4. Since the authors have claimed to be used as a protective layer in PV devices. It is recommended to include a water contact angle (as reported; https://www.sciencedirect.com/science/article/abs/pii/S0040609022003984) on the sputtered film. It is suggested to discuss the surface properties compared with other kinds of sputtered films.

We express our thanks to the reviewer for the excellent idea of including a water contact angle test. These tests were performed by our team and water contact angle results were added to the manuscript (Figs. 11 and 12).

Regarding a possible comparison to other kinds of sputtered films, this would be a difficult task. To our knowledge there are no results in literature for ultrathin films of CrN deposited on Si substrate with the two methods used in the paper that use such short deposition times, usually deposition times greater than 30 minutes are used. Besides this, to our knowledge, no other paper explores the quantitative characterization by fractal dimension, correlation, and energy of CrN films deposited on Si substrate with the two methods presented in the paper.

5. It is suggested to include an error bar in the tabulated results wherever it applies. 

Error bars were added in the tabulated results according to the reviewer’s suggestion.

Reviewer 4 Report

This paper needs a thorough revision.

(1) This paper is more like a experimental report than a scientific research paper.

(2) The abstract part should be re-written to tell the readers what is the most important thind for this work.

(3) The introduction part also needs to be improved. The reviewer can not find what is new, what is the problem, what is the meaning of this work.

(4) A shematic of the setup of the experiment of depositing coating is necessary in the experimental part.

(5) The information of the euipments used in this paper should be provided.

(6) What is the test item for in this research work, to support what kind of conclusion?

(7) Use Figure to express the table data.

(8)The quality of the cross-sectional morphologies are very poor.

(9) There are some other grammar errors. 

Author Response

We address many thanks to the Reviewer for their thoughtful and professional insights in respect to our manuscript. For an easy-to-read answer, we will respond point-by point in the following. Their comments are left as they were, while our responses are inserted correspondingly with green color. 

• This paper is more like an experimental report than a scientific research paper.

Yes, the reviewer is correct, the article focuses on the experimental results, which we consider increasing the manuscript scientific value.   

• The abstract part should be re-written to tell the readers what the most important thing for this work is.

The abstract has been re-written to tell the readers what the most important of this work is.

• The introduction part also needs to be improved. The reviewer cannot find what is new, what is the problem, what is the meaning of this work.

According to the valuable reviewer’s suggestion, We have improved the introduction section in        order to explain the meaning of our work.

What is new about this study is the characterization of DC and HiPIMS sputtered deposition that are on the order of tenths of nanometers. There are no existing results in literature for such thin films of CrN deposited on Si substrates using the two deposition methods presented in this paper. Particularly no paper explores the quantitative characterization by fractal dimension, correlation, and energy.

• A shematic of the setup of the experiment of depositing coating is necessary in the experimental part.

A schematic of the experiment of deposition coating is not necessary, as the apparatus is not one that the authors build (they are not home-made equipment). The type and producers of the equipment are clearly expressed in the manuscript and details about them may be extracted directly from the producers.  

• The information of the equipements used in this paper should be provided.

All the information about the equipment used in the paper has been added in the materials and methods part.

• What is the test item for in this research work, to support what kind of conclusion?

The test items for this paper is the study of the properties of ultrathin CrN coatings on Si substrate deposited using direct current sputtering (DC) and high-power impulse magnetron sputtering (HiPIMS)

• Use Figure to express the table data.

The use of figures to express table data is not necessary as there would only be three data points for each sample. The reviewer is correct, figures for the table data would be useful for the visualization of trends if we would have more data points.

       (8) The quality of the cross-sectional morphologies are very poor.

The cross-sectional morphologies have been revised with ones that have identical appearance/thickness of Si-substrate. So that readers can clearly see the variation of CrN thickness only.

Round 2

Reviewer 1 Report

The Authors have duly taken into account all criticisms listed in my previous report. Although changes are not very comprehensive and some of the statements reported in the response letter are questionable (in particular those relating with the ability of AFM and SEM in reconstructing morphologies), the paper is improved in terms of completeness and quality of presentation. This enhances In turn the scientific soundness to an (almost) adequate level.

Therefore, in my opinion the manuscript can be accepted for publication in the present form.

Author Response

We thank the reviewer for their useful advice. As we wrote in the previous answer, in the article we discuss two separate measurements: the
topography obtained with AFM and the morphology obtained with SEM. In addition, the two measurements are made on very different scales, and similarities between the two measurements cannot be highlighted

Reviewer 2 Report

The authors have addressed most of my comments, but the following should also be addressed:

-The reasoning for the deposition times must be explicitly explained in the text;

Author Response

-The reasoning for the deposition times must be explicitly explained in the text;

We have updated the text to provide an explanation for the chosen deposition times

The deposition times of 5, 10, and 15 minutes were chosen specifically because we wanted the deposited layer to be below 400 nm thickness. During our experiments we tested different deposition times and noticed that for deposition times greater than 15 minutes, the thickness of the deposited layer would exceed 400 nm. For the research purpose of studying depositions with thickness lower than 400 nm we chose 15 minutes as the maximum deposition time because for this deposition time we would obtain samples with the thickness of the deposited layer near 400 nm. We also chose to produce samples in which the deposited layer is thinner by using the lower deposition times of 5 and 10 minutes.

Reviewer 3 Report

Accept

Author Response

We thank the reviewer for their useful advice

Reviewer 4 Report

（1）The only informative sentence I see in th abstract is "Our results show that using the magnetron techniques, ultrathin CrN films with excellent me chanical properties can be obtained".  Give more data supported conclusions in the abstract.

(2) The reviewer cannot find what is important in this paper. What kind of problem is to be sovled in this paper? What is the purpose?

Author Response

（1）The only informative sentence I see in th abstract is "Our results show that using the magnetron techniques, ultrathin CrN films with excellent me chanical properties can be obtained".  Give more data supported conclusions in the abstract.

We have updated the abstract to provide more data supported conclusions, mainly releated to Young’s modulus and stiffness

New Abstract:  Chromium Nitride thin films are known for their good mechanical properties. We present the characteristics of ultrathin Chromium Nitride films under 400 nm thickness deposited on silicon substrates by direct current and high-power impulse magnetron sputtering techniques. The methods of investigation of the CrN films were scanning electron microscopy, atomic force microscopy, and nanoindentation. Qualitative and quantitative analysis were performed using the AFM and SEM images by fractal dimension, surface roughness and gray-level co-occurrence matrix methods. Our results show that using the magnetron techniques, ultrathin CrN films with excellent mechanical properties were be obtained, characterized by values of Young’s modulus between 140 GPa and 250 GPa for the samples obtained using high-power impulse magneton sputtering (HiPIMS) and between 240 GPa and 370 GPa for the samples obtained using direct current sputtering (DC). Stiffness measurements also reveal the excellent mechanical properties of the investigated samples, the samples obtained using HiPIMS sputtering have stiffness values between 125 N/m and 132 N/m and the samples obtained using DC sputtering have stiffness values between 110 N/m and 119 N/m

(2) The reviewer cannot find what is important in this paper. What kind of problem is to be sovled in this paper? What is the purpose?

What is important in this paper is to obtain ultrathin CrN films deposited on Si substrate using HiPIMS and DC sputtering techniques. The purpose of obtaining these ultrathin CrN depositions is related to the substrate on which they are deposited, in our case Si. The main purpose is to protect the Si substrate, so it is necessary to have a deposition with very good mechanical properties. In addition, for the use in the photovoltaic applications, these depositions must also be transparent. This is achieved with such ultrathin depositions as measured using Young’s modulus and Stiffness values. These depositions can also have a role in the protection of glass surfaces